Skip to main content

Venue, Counterparty & Curator Risk

The strategy trades on venues we do not control, prices itself from data we do not produce, and settles through infrastructure we operate. Each is a dependency.

Venue risk​

Execution happens on Hyperliquid. A failure, exploit, halt or insolvency there affects the strategy's ability to trade or withdraw.

Concretely, a venue can halt trading or withdrawals, socialise losses across users after a large liquidation event, suffer an exploit that removes collateral, or change margin and funding rules in a way that forces a position change at a bad moment.

⚠️ Execution is concentrated on one venue. The live strategy trades overwhelmingly on Hyperliquid, and there is currently little venue diversification to absorb an event there.

Liquidation​

Perpetual positions are margined. If margin requirements are breached the venue can liquidate at a price the strategy does not choose.

The gross notional cap, full collateralisation and volatility targeting keep the portfolio well inside its margin requirement in normal conditions. A large gap move against a concentrated position is where they are not enough.

Liquidity​

Large positions can be difficult to exit at modelled prices in thin conditions. The strategy models impact per trade, but conditions change faster than a model recalibrates.

Meme markets are liquid in normal conditions and can become one-sided quickly. Tokenized RWA markets are structurally thinner than the major crypto perpetuals, which is the trade-off that comes with the low correlation they supply.

Redemption​

Redemption is onchain and does not require our permission. Two things still delay it: the 7 day cooldown, and settlement, which needs a NAV push from the valuation manager.

If our backend is down, requests queue. The position is not lost and the assets stay in the vault, but the exit is not immediate.

⚠️ If you need same-day liquidity, this is not the product.

Oracle and pricing​

NAV is computed off-chain from venue state and pushed onchain. If that computation is wrong, shares are minted or burned at the wrong price, transferring value between whoever transacts in that window and everyone else.

The guards on Vault Protections bound the size of that error. They do not eliminate it.

Curator and operational risk​

The signal engine, orchestrator, data pipelines and curator service can fail. We run monitoring with SMS escalation for a small set of critical conditions.

That monitoring exists because of a real incident: in June 2026 a signing key ran out of gas and settlements froze for 34.7 hours while a single muted alert was deduplicated silently.

A backend outage results in a missed rebalance and delayed settlement. A process that is dead sends no alerts.

Custody configurations differ between deployments and determine what a compromised operational key can reach. See Security, and read the Safe configuration onchain rather than relying on a documentation page.